TL;DR: Cohort retention metrics reveal whether your SaaS business is getting better or worse at keeping customers. Most companies only track aggregate churn, which averages together customers who behave completely differently and hides trends until they’re catastrophic. Proper cohort retention analysis groups customers by signup period, tracks their behavior month by month, and lets you compare how different groups perform. This shows whether product improvements are working, which acquisition channels bring stickier customers, and where revenue growth will come from before it shows up in your P&L.
We talk to SaaS founders monthly who are confident about their retention. They track overall churn at 3% monthly, it’s been stable for six quarters, and the board accepts it as reality.
Then we build cohort tables and everything changes. The January 2024 cohort is at 88% retention after six months. The July 2024 cohort is at 79% retention after six months. Retention is degrading 1.5% monthly, but aggregate churn looks stable because the company is growing fast enough that new customers mask the deterioration.
This is the fundamental problem with aggregate metrics. They blend together customers from different time periods, different acquisition sources, different product experiences, and produce a single number that tells you almost nothing about business trajectory.
Cohort retention analysis solves this by tracking groups of customers who share a common characteristic (usually signup month) and comparing their behavior over time. This reveals trends that aggregate numbers hide completely.
Build your cohort analysis around three fundamental retention metrics, each telling a different story about customer behavior.
Logo retention measures what percentage of customers remain active. If you start January with 100 customers and end December with 85 of those same customers still active, that’s 85% 12-month logo retention. This tells you if customers are finding enough value to stick around.
Revenue retention measures what percentage of revenue remains from a cohort. Those 85 customers might be paying 105% of what they paid originally if they’ve expanded. Logo retention of 85% with revenue retention of 105% means you’re losing customers but the survivors are growing enough to more than offset the losses.
Net revenue retention combines retention and expansion into a single metric. Take starting MRR from a cohort, add expansion MRR, subtract churn MRR and contraction MRR, divide by starting MRR. Above 100% means the cohort is growing in value despite losing some customers. Above 110% is good, above 120% is exceptional.
Track all three because they tell different stories. A company with 90% logo retention and 95% revenue retention is losing customers who aren’t expanding. A company with 80% logo retention and 110% revenue retention is losing customers but the keepers are expanding aggressively. These require completely different strategic responses.
The standard cohort retention table puts cohorts (usually by signup month) in rows and months-since-signup in columns. Here’s what it looks like:
Cohort | M0 | M1 | M3 | M6 | M12 | M24
Jan 2023 | 100% | 94% | 88% | 84% | 79% | 72%
Apr 2023 | 100% | 95% | 89% | 85% | 81% | –
Jul 2023 | 100% | 93% | 87% | 82% | 78% | –
Oct 2023 | 100% | 92% | 85% | 80% | – | –
Jan 2024 | 100% | 91% | 84% | 79% | – | –
Read this table two ways. Horizontally shows individual cohort behavior over time. The January 2023 cohort retained 94% after one month, 88% after three months, 79% after twelve months. Vertically compares cohorts at equivalent points in their lifecycle. At month 6, January 2023 was at 84% while January 2024 is at 79%, suggesting retention is degrading.
This pattern is critical. When newer cohorts show worse retention than older cohorts at the same lifecycle stage, something changed for the worse. Maybe product quality declined, maybe you started targeting different customers, maybe onboarding broke. The cohort table shows the symptom immediately even if you don’t know the cause yet.
Build separate tables for logo retention and revenue retention. Often these diverge meaningfully. Logo retention might be declining while revenue retention improves because you’re getting better at expanding the customers who stay.
Update these tables monthly. Review them in management meetings alongside acquisition metrics. When cohort retention degrades, investigate that week, not next quarter.
Don’t stop at time-based cohorts. Segment by any attribute that might affect retention and build separate cohort tables for each.
Segment by acquisition channel. We worked with a company whose organic search cohorts had 92% six-month retention while paid search cohorts had 76% six-month retention. Same product, same price, dramatically different retention. They shifted budget from paid to organic and content, improving blended retention by 8 percentage points.
Segment by customer size (MRR bands). Small customers typically churn faster than large ones. A company might see 70% 12-month retention for customers under $100 MRR, 85% for customers between $100-500 MRR, and 94% for customers over $500 MRR. This informs pricing strategy and what customer segments to target.
Segment by product tier or package. Do customers on your premium plan retain better than those on basic? If premium has 95% retention while basic has 80% retention, that’s evidence that higher-value customers stick around. Build features that drive upgrades.
Segment by industry vertical if you serve multiple industries. Maybe healthcare customers retain at 88% while retail customers retain at 72%. This might mean you have better product-market fit in healthcare or that retail customers face different budget pressures.
Segment by sales motion. Self-serve signups versus demo-driven sales versus enterprise deals often have completely different retention profiles. Self-serve might have high early churn but customers who survive month 3 are sticky. Enterprise deals might have near-zero early churn but need renewal management at month 12.
Build as many segmented cohort tables as you can maintain. Each reveals patterns that help you allocate resources better, target more effectively, and forecast more accurately.
Plot cohort retention as curves rather than tables and you’ll see patterns that numbers hide. A healthy B2B SaaS retention curve typically shows high early churn (months 1-3) as bad-fit customers self-select out, then flattening as you reach customers getting real value.
The curve shape tells you everything about business viability:
Fast decline that flattens: 95% month 1, 88% month 3, 84% month 6, 81% month 12, 79% month 24. This is healthy. Early churn clears out poor fits, then retention stabilizes because remaining customers have found value.
Linear decay that never flattens: 95% month 1, 90% month 3, 85% month 6, 80% month 12, 75% month 24. This is crisis. Customers keep leaving at constant rates forever, suggesting fundamental product-market fit problems. You don’t have a retention problem, you have a value problem.
Secondary churn spikes: 95% month 1, 91% month 3, 89% month 6, 85% month 12 (ouch), 82% month 24. Something happens at month 12 that causes abnormal churn. Maybe annual contract renewals, maybe onboarding effects wear off, maybe competitors target customers after they’ve been around a year.
Improving curves over time: April 2023 cohort is at 79% month 12, October 2023 cohort is at 83% month 12, April 2024 cohort is tracking toward 86% month 12. This is validation that you’re improving. Product is better, onboarding is better, or targeting is better.
Plot curves for your last 8-12 cohorts and look for these patterns. The shape reveals whether you’re building a viable business or fighting gravity.
Cohort retention data transforms revenue forecasting from guesswork into math. Instead of assuming “we’ll grow 30% annually,” you model exactly what happens to each cohort based on historical retention patterns.
Start with your existing customer cohorts. Apply historical retention curves to project their future revenue. The January 2024 cohort that started at $50K MRR will probably follow the same retention curve as previous cohorts, so you can project their MRR in months 12, 18, and 24.
Add planned new customer acquisition. If you’re adding $40K in new customer MRR monthly, apply your typical retention curves to those future cohorts. First-month customers will probably retain at 94%, three-month at 89%, and so on.
Sum across all cohorts to get total projected MRR. This bottoms-up approach is dramatically more accurate than trend-based projections because it accounts for cohort aging and composition.
We built a forecast for a client showing they’d miss their 100% growth target despite doubling new acquisition. Why? Existing customer decay offset new revenue more than they expected. Historical cohorts were rolling off at 2.5% monthly while new cohorts started small and took time to compound. The cohort-based model showed realistic 65% growth, which changed their hiring and spending plans dramatically.
When we analyze retention cohorts for clients, patterns emerge that explain growth problems founders couldn’t understand.
Retention degrading by cohort: Newer customers churning faster than older ones. This usually means product-market fit is slipping, you’re acquiring worse-fit customers, or onboarding broke. Fix it immediately because compound effects are brutal.
Retention improving by cohort: Newer customers sticking better than older ones. This validates that improvements are working. Maybe you fixed bugs, improved onboarding, or refined targeting. Double down on what’s working.
Channel-specific retention gaps: Some acquisition sources bring customers who churn at 2x the rate of others. Stop spending on the bad channels even if they seem cheaper. A customer who churns in month 3 has negative lifetime value after acquisition costs.
Early churn spikes: Losing 15-20% of customers in month 1. This signals onboarding problems, expectations mismatch from sales process, or activation issues. Focus product resources on first-month experience.
Anniversary churn spikes: Losing abnormal percentages at month 12, 24, or 36. This suggests contract renewal issues or lack of ongoing value communication. Build renewal motions and customer success programs around these inflection points.
Expansion timing: Cohorts that stick past month 6 expand at 4% monthly while cohorts under 6 months don’t expand. This tells you when to introduce upsells and expansion conversations.
Companies that master cohort retention build monthly dashboards that leadership teams actually use:
Primary retention table showing last 12 cohorts with logo retention at M1, M3, M6, M12
Revenue retention table for the same cohorts
Net revenue retention trend line showing whether overall retention is improving or degrading
Segmented views by channel, customer size, and product tier
Retention curve plots showing whether newer cohorts are tracking better or worse than historical cohorts
Alert indicators when any cohort shows retention 10% worse than historical averages at equivalent lifecycle stage
Review this dashboard monthly alongside acquisition metrics. When retention degrades, investigate immediately. When retention improves, understand why so you can replicate it.
The dashboard turns retention from a lagging indicator (you notice problems after they hurt revenue) into a leading indicator (you see warning signs months before revenue impact).
—
Q: How many months of data do we need before cohort analysis is meaningful?
You need at least 6 months of cohorts with at least 3 months of retention history on the newest cohort before patterns become clear. With only 2-3 months of data, random variation swamps real signals. Once you have 12+ months of cohorts, patterns become reliable and you can confidently forecast based on historical retention curves. Early-stage companies should start tracking immediately even with limited data, but don’t over-interpret early results. The value compounds as you accumulate history.
Q: What’s a “good” retention rate for SaaS companies?
It depends entirely on your market and business model. SMB monthly contracts typically see 5-8% monthly logo churn (60-70% annual retention), mid-market annual contracts see 15-25% annual churn (75-85% retention), enterprise contracts see 5-15% annual churn (85-95% retention). More important than absolute retention is the trend (improving or degrading?) and net revenue retention (is expansion offsetting churn?). A company with 80% logo retention but 110% net revenue retention is healthier than one with 90% logo retention and 90% net revenue retention.
Q: How do we improve cohort retention once we’ve identified problems?
Start by segmenting to isolate the issue. Is retention bad across all cohorts or specific to certain channels, customer sizes, or time periods? Then investigate root causes through customer interviews, cancellation surveys, and usage analysis. Common fixes include improving onboarding for early churn, building expansion paths for stagnant cohorts, adding missing features that customers request before churning, or stopping acquisition from channels that bring poor-fit customers. The key is acting fast—every month of degraded retention compounds into lost revenue for years.